I -and anyone literate for that matter- can tell you the exact composition of the atmosphere on Mars: 96% carbon dioxide, 1.9% argon, 1.9% nitrogen, and traces of free oxygen, carbon monoxide, water and methane, among other gases.
Neither I nor anyone else though can tell you which way the Britons will vote in the upcoming Brexit referendum. We know amazing facts about things way far away from ourselves but we do not know the first thing about our backyard or neighborhood.
Years of scarce computing capacities have forced us to learn how to sample a large unmanageable observation universe in order to be able to compute measurements on a smaller manageable sample from which to extrapolate on the universe. Nonetheless we often fail spectacularly and we often do so exactly when it matters the most. When we do fail we believe that our sample was skewed (or that the polled sample was refusing to reveal its true preferences). So what are we to make out of the fact that the Economist currently has the remain camp up by one percentage point, especially when a stubbornly persisting 11% registers at “don’t know”. And what are we to conclude when the Financial Times poll of polls has the leave camp currently up by 1% with 11% registering under “undecided”?
A dispassionate and statistically literate observer might say that the diversity prediction theorem implies the average of all polls (i.e. the Financial Times poll) is better than any other single poll alone knowing full well however that being better than awful is not gonna cut it. So if you ask me polls do not help at this juncture not so much because the sample may be biased or the contractor of the poll might want to see a certain outcome but because 10% or more of the polled Britons do not know which way they will go or refuse to reveal what they will do and those that do are likely to say one thing and vote another.
Then we have the bookies whose “credibility” rests on the fact that they put their money where their mouth is but often the amounts they bet are negligible and they have been known to fail spectacularly on occasion, some of them offering odds of 5,000 to 1, early in the previous soccer season, for Ranieri’s Leicester to win the Premiere league.
Finally we have the real thing, the voters themselves who think, feel, argue, agonize and some, unfortunately, may kill or die. Some want to know which way to go, some feel one way and think another, some may hold a conviction and try to recruit more under their flag . They seek for information, arguments, they try to find others like themselves, some want to know how bad it will be if the “others” win. Where are we to get any kind of reading about what is really happening in the UK electorate just one day before voting starts?
There is no better place than Google Search: the demand for one kind of information or the other reveals facts about the state of the individuals and search is the act of expressing that demand. If we are lucky then searches containing the word leave are related to intend to vote leave and those with stay are related to intent to vote remain. Such searches may have some noise of course coming from such things as “maternity leave” or “why did so and so leave big brother” and some may be related to music than to the referendum. With some luck and adjustment we may be able to 1. clean the searches of irrelevant ones and 2. argue that they reveal voter intent. I am confident to mostly have achieved the first and I can see that many -but not all- of the top searches are indeed related to voting intent. For example if you are searching for “remain poster” you are likely to sympathize with the remain side and searches for “leave campaign” may be because you want to join it or read about it to feel convinced you are doing the right thing sympathizing with the leave side. So modulo that I am not fully convinced about the validity of my data it is the best I can currently get and I say that the remain side is still ahead but that the brexiteers are catching up in a hurry, after their side lost a lot of traction in the aftermath of the tragic shooting death of Jo Cox.